We propose computational approaches aiding automated compilation of molecular networks from research literature, cleansing of the resulting database, and assessing reliability of facts stored in the database.